Object Detection

TFLite Runtime

No. Model Name Description

1 Yolo-V5 Real-time object detection using anchor-based detection heads.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolov5_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov5_float.tflite            <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov5_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME settings="{\"confidence\": 70.0}" ! text/x-raw ! queue ! obj_mux.

2 Person-Foot-Detection Specialized foot and person localization for tracking pipelines.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	person_foot_detection_float.tflite
foot_track_net.json	foot_track_net.json
foot_track_net_settings.json	foot_track_net_settings.json
ai_demo_sample.mp4	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp person_foot_detection_float.tflite       <user>@<device-ip>:$HOME/models/
scp foot_track_net.json                      <user>@<device-ip>:$HOME/labels/
scp foot_track_net_settings.json             <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4                            <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=person_foot_detection_float.tflite
export LABELS_NAME_1=foot_track_net.json
export LABELS_NAME_2=foot_track_net_settings.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=qpd labels=$HOME/labels/$LABELS_NAME_1 settings=$HOME/labels/$LABELS_NAME_2 ! text/x-raw ! queue ! obj_mux.

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	person_foot_detection_w8a8.tflite
foot_track_net.json	foot_track_net.json
foot_track_net_settings.json	foot_track_net_settings.json
ai_demo_sample.mp4	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp person_foot_detection_w8a8.tflite        <user>@<device-ip>:$HOME/models/
scp foot_track_net.json                      <user>@<device-ip>:$HOME/labels/
scp foot_track_net_settings.json             <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4                            <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=person_foot_detection_w8a8.tflite
export LABELS_NAME_1=foot_track_net.json
export LABELS_NAME_2=foot_track_net_settings.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=qpd labels=$HOME/labels/$LABELS_NAME_1 settings=$HOME/labels/$LABELS_NAME_2 ! text/x-raw ! queue ! obj_mux.

3 Yolo-X Anchor-free YOLO variant with decoupled detection head.

Steps to Run on QIMSDK

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	yolox_w8a8.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolox_w8a8.tflite              <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolox_w8a8.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

4 YOLOv11-Detection Latest YOLO generation with enhanced feature extraction.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolov11_det_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov11_det_float.tflite       <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov11_det_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	yolov11_det_w8a8.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov11_det_w8a8.tflite        <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov11_det_w8a8.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

5 Yolo-v7 High-accuracy detection with re-parameterizable convolutions.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolov7_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov7_float.tflite                 <user>@<device-ip>:$HOME/models/
scp yolov8.json                         <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4           <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov7_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	yolov7_w8a8.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov7_w8a8.tflite             <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov7_w8a8.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

6 YOLOv8-Detection Streamlined single-stage detector with CSP bottleneck backbone.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolov8_det_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov8_det_float.tflite        <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov8_det_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	yolov8_det_w8a8.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov8_det_w8a8.tflite         <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov8_det_w8a8.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

7 YOLOv10-Detection NMS-free dual-label assignment for end-to-end detection.

Steps to Run on QIMSDK

🟢 W8A8 — Model Precision

Download Required Files

File	Save as
Model	yolov10_detection_w8a8.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov10_detection_w8a8.tflite  <user>@<device-ip>:$HOME/models/
scp yolov8.json                    <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4      <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov10_detection_w8a8.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolov10_detection_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolov10_detection_float.tflite  <user>@<device-ip>:$HOME/models/
scp yolov8.json                     <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4       <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolov10_detection_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

8 DETR-ResNet101 Transformer-based detector with deeper ResNet-101 backbone.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	detr_resnet101_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp detr_resnet101_float.tflite  <user>@<device-ip>:$HOME/models/
scp yolov8.json                  <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4    <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=detr_resnet101_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME settings="{\"confidence\": 70.0}" ! text/x-raw ! queue ! obj_mux.

9 DETR-ResNet50 Transformer-based detector with ResNet-50 backbone.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	detr_resnet50_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp detr_resnet50_float.tflite  <user>@<device-ip>:$HOME/models/
scp yolov8.json                 <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4   <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=detr_resnet50_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

10 Conditional-DETR-ResNet50 Conditional cross-attention DETR for faster convergence.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	conditional_detr_resnet50_float.tflite
coco_labels.json	coco_labels.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp conditional_detr_resnet50_float.tflite  <user>@<device-ip>:$HOME/models/
scp coco_labels.json                        <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4               <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=conditional_detr_resnet50_float.tflite
export LABELS_NAME=coco_labels.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME settings="{\"confidence\": 50.0}" ! text/x-raw ! queue ! obj_mux.

11 RF-DETR Region-aware feature DETR with improved small-object detection.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	rf_detr_float.tflite
coco_labels.json	coco_labels.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp rf_detr_float.tflite            <user>@<device-ip>:$HOME/models/
scp coco_labels.json                <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4       <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=rf_detr_float.tflite
export LABELS_NAME=coco_labels.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME settings="{\"confidence\": 70.0}" ! text/x-raw ! queue ! obj_mux.

12 RTMDet Real-time instance-level detector with dynamic label assignment.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	rtmdet_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp rtmdet_float.tflite        <user>@<device-ip>:$HOME/models/
scp yolov8.json                <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4  <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=rtmdet_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=external external-delegate-path=libQnnTFLiteDelegate.so external-delegate-options="QNNExternalDelegate,backend_type=htp,log_level=(string)1;" ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME settings="{\"confidence\": 50.0}" ! text/x-raw ! queue ! obj_mux.

13 Yolo-R Multi-scale object detection with implicit knowledge distillation.

Steps to Run on QIMSDK

🟡 Float32 - Model Precision

Download Required Files

File	Save as
Model	yolor_float.tflite
yolov8.json	yolov8.json
Input video	ai_demo_sample.mp4

If any downloaded file is a .zip archive, extract it on your host machine before copying: unzip filename.zip

Copy Files to Device

# Replace $HOME to the appropriate device path before running the commands.
# For QLI:    /root
# For Ubuntu: /home/ubuntu
# Modify this based on your platform and ensure files are copied to the correct location on the device.

ssh <user>@<device-ip> "mkdir -p $HOME/{models,labels,media,media/output}"
scp yolor_float.tflite         <user>@<device-ip>:$HOME/models/
scp yolov8.json                <user>@<device-ip>:$HOME/labels/
scp ai_demo_sample.mp4  <user>@<device-ip>:$HOME/media/

Connect to device

# Run from your host machine — replace <user> and <device-ip>
ssh <user>@<device-ip>

Set Environment Variables

Qualcomm Linux

export MODEL_NAME=yolor_float.tflite
export LABELS_NAME=yolov8.json
export SRC_VIDEO_NAME=ai_demo_sample.mp4

Run the Pipeline

gst-launch-1.0 -e --gst-debug=2 \
filesrc location=$HOME/media/$SRC_VIDEO_NAME ! qtdemux ! h264parse ! \
v4l2h264dec capture-io-mode=4 output-io-mode=4 ! video/x-raw,format=NV12 ! queue ! \
tee name=t ! qtimetamux name=obj_mux ! qtivoverlay ! waylandsink fullscreen=true sync=true \
t. ! queue ! qtimlvconverter ! queue ! \
qtimltflite model=$HOME/models/$MODEL_NAME delegate=gpu ! queue ! \
qtimlpostprocess module=yolov8 labels=$HOME/labels/$LABELS_NAME ! text/x-raw ! queue ! obj_mux.

Blog

Overview

Getting Started

Building AI Pipelines

Examples

Reference

Advanced Topics